Web to app funnels and the last step

W2A funnels are measured obsessively at every step except the one where the money leaves.

6 August 2026 by PAD team

The typical funnel

Ad, landing page, a few quiz questions, a personalised result, then a Download button pointing at the App Store. Every step is instrumented, drop-off is monitored per question, and the last transition is usually a black box.

Why the last step is invisible

Your analytics records the button click. The ad platform records the original click. Neither can see whether the App Store actually opened, because that happens outside the page.

So a funnel that loses a fifth of its traffic at the final tap looks like a funnel with a modest install rate, and the optimisation effort goes into the quiz.

What breaks

The whole funnel runs inside the in-app browser of Instagram, Facebook or TikTok. Pages render fine. Then the Download button points at apps.apple.com, Apple answers with the itms-appss:// scheme, and the WebView cannot follow it.

The visitor has answered eight questions, seen their result, and reached for the app. That is the most expensive moment in the funnel to lose someone.

Finding the leak

• Run the funnel end to end from inside Instagram on a real iPhone
• Compare the last-step conversion between iOS and Android, since Android rarely fails this way
• Instrument the button to report whether the escape succeeded, not just that it was pressed
• Compare traffic-objective funnels with install campaigns for the same app

The fix is one link

Nothing about the funnel needs to change. The Download button stops pointing at the store directly and points at a routing link instead, which escapes the browser and then goes to the store, carrying your campaign parameters.

If you use a tracker, point the escape at the tracker URL and your attribution continues working exactly as before.